A few weeks ago I posted a rather desperate plea for help on SGMLS'
unwillingness to accept TEI documents. After seeking help locally I
originally thought the problem might be solved by "compiling" the TEI
DTD for our specific requirements. Thank you, readers of TEI-L, for
disabusing me of that notion.
What turned out to be the chief cause of my falling out with SGMLS was
the SGML declaration. I confess that I was not aware of its vital
importance. P3 is very coy about it, devoting a single paragraph to it;
Van Herwijnen mentions it briefly in 3.1.1, not to return to it till
Part III, where he discusses it in a way which, frankly, baffles me.
I was alerted to the importance of the SGML declaration by some helpful
postings (thank you, Dominic Dunlop, Richard Light and Tobias Rischer).
When I tried to add the TEI one (TEISGML.DEC) to the command line, this
caused SGMLS to spew enough error messages to lead me to think at first
that it did not make any difference at all. (In fact, on closer
inspection, it did: the size of the error message file went down from
almost 0.5Mb to some 25k. Still a respectable size, but a decided
improvement.)
What turned out to be the case was that for some reason the GRPCNT in
the TEISGML.DEC file is set too low: 64 instead of something like
120-200. Most of the parser's anger resulted from this parsimonious
setting.
What with this inbuilt stumbling block, the lack of attention to the
SGML declaration in P3 and the incomprehensible account of it in Van
Herwijnen, I suppose it was rather easy to be led astray here. My
obvious question is, why on earth is the incompatible setting included
in a file that goes with the full TEI dtd distribution? (I can't
remember where I found it, but the file was called P3DTDs.tar.)
Incidentally, also many thanks for recommending NSGMLS, which is a
decided improvement for the amateur user: it even parses a TEI document
without any SGML declaration. A look at the parser's default SGML
declaration will show why. GRPCNT is set to a whopping 99999999; not
even the TEI appears to want that much.
Thank you also, Arjan Loeffen, Nick Summers, Alfie Kirkpatrick, Harry
Gaylord, Fotis Jannidis.
Adriaan van der Weel
Book and Publishing Studies
Dept of English, University of Leiden
P.O. Box 9515
2300 RA Leiden
Netherlands
Phone +31 71 5272144; fax 5272615
[log in to unmask]
|